pith. sign in

arxiv: 2606.07545 · v1 · pith:KIYBM544new · submitted 2026-05-02 · 💻 cs.CY

Reshaping Undergraduate Computer Science Education in the Generative AI Era

Yi-Chieh Lee , Nattapat Boonprakong , Yugin Tan , Harold Soh , Alex Potanin , Viraj Kumar , Anoop K. Sinha , Chen Qian
show 15 more authors
Paul Denny Mennatallah El-Assady Ian Oakley Jake Renzella Amy Zhang Jat Singh Wee Sun Lee Hsuan-Tien Lin Jane L. E Anthony Tang Margaret M. Burnett Sowmya Somanath Renwen Zhang Vicky Charisi Alexandra I. Cristea
This is my paper

Pith reviewed 2026-07-01 00:52 UTC · model grok-4.3

classification 💻 cs.CY
keywords computer science educationgenerative AIcurriculum reformAI-generated artifactssystem designcritical evaluationundergraduate teachingAI-native competencies
0
0 comments X

The pith

CS curricula should shift from teaching implementation skills to understanding and verifying AI-generated artifacts.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Generative AI now automates many tasks such as coding, debugging, and basic software design that have defined undergraduate computer science education for decades. The paper claims this automation requires a reorientation toward skills for system design, abstraction, and critical evaluation of AI outputs to keep graduates effective. Two international workshops with faculty, industry practitioners, and students supplied the evidence for which skills remain essential and which are declining in importance. The authors propose embedding these priorities as incremental breadcrumbs of exercises and nudges inside existing courses rather than requiring wholesale replacement. If the claim holds, graduates would become capable of co-creating and managing AI artifacts instead of performing the automated work themselves.

Core claim

The central claim is that to prepare future computer science graduates for creating, solving problems, and co-creating with AI, curricula must foster AI-native competencies, re-center fundamental education on abstraction and design, enhance advanced pathways, embrace new pedagogies, and shift institutional support, all derived from workshop identification of preserved versus less important skills.

What carries the argument

The breadcrumbs approach of embedding helpful nudges and engaging exercises for new competencies directly into the current curriculum structure.

If this is right

  • Students will practice verifying and refining AI outputs as a core activity rather than writing code from scratch.
  • System design and abstraction will receive more instructional time across multiple courses.
  • Incremental updates can occur inside existing course sequences without waiting for full program redesign.
  • Institutional policies will need to support new forms of assessment and faculty preparation for AI-integrated teaching.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Assessment methods may need to evolve to test students' ability to critique and improve AI artifacts rather than produce original code.
  • Parallel adjustments could appear in adjacent fields such as information systems or software engineering programs.
  • Early experiments could test whether short modules on AI output evaluation improve student outcomes in existing programming courses.

Load-bearing premise

The workshops produced representative and actionable insights about which skills will remain important for future graduates.

What would settle it

A controlled comparison of job performance or problem-solving outcomes between cohorts trained under the proposed verification-focused curriculum versus traditional implementation-focused training would show no advantage or a reversal of the predicted benefit.

Figures

Figures reproduced from arXiv: 2606.07545 by Alexandra I. Cristea, Alex Potanin, Amy Zhang, Anoop K. Sinha, Anthony Tang, Chen Qian, Harold Soh, Hsuan-Tien Lin, Ian Oakley, Jake Renzella, Jane L. E, Jat Singh, Margaret M. Burnett, Mennatallah El-Assady, Nattapat Boonprakong, Paul Denny, Renwen Zhang, Sowmya Somanath, Vicky Charisi, Viraj Kumar, Wee Sun Lee, Yi-Chieh Lee, Yugin Tan.

Figure 1
Figure 1. Figure 1: Our workshops are initiatives by the AI4SG Lab, at the NUS School of Computing, and [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Photo from Day 2. The faculty workshop features multiple rounds of group brainstorming [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Faculty participants proposed the Breadcrumb strategy as a lightweight solution to foster [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Example brainstorming work from the student workshop. Students voiced opportunities, [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Criteria for CS Education Reforming Solutions [PITH_FULL_IMAGE:figures/full_fig_p021_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Undergraduate CS curriculum changes must involve recentering fundamental education [PITH_FULL_IMAGE:figures/full_fig_p025_6.png] view at source ↗
read the original abstract

Generative AI represents a turning point for Computer Science (CS) education. In recent decades, post-secondary CS education has largely focused on what has been seen as practical software engineering skills: implementation-level programming, debugging, testing, and software design, analysis, and documentation. However, this framing is becoming less tenable as generative AI automates many of these tasks, challenging their centrality in CS education. To keep pace with advances in AI technology, CS curricula should consider a shift toward understanding and verifying AI-generated artifacts. This white paper outlines the findings of two international NUS-Google Workshops in Singapore, where we convened faculty members, industry practitioners, and students, and proposes a strategic response to reshape how CS should be taught at the undergraduate level. Based on the findings, we identify critical skills that must be preserved and those that are becoming less important. By incorporating these skills as "breadcrumbs," we can provide helpful nudges and engaging exercises within the current curriculum, enhancing learning experiences for everyone. We believe that to effectively prepare future computer science graduates, capable of creating, solving problems, and managing, as well as co-creating, artifacts with AI. It is important to consider a shift in curricula. Emphasizing system design, abstraction, and critical evaluation could greatly enhance their education and readiness for the challenges ahead. We propose prerequisites for solutions to reform CS education by fostering AI-native competencies, re-centering fundamental education, enhancing advanced pathways, embracing new pedagogies, and shifting institutional support.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims that generative AI is automating core implementation, debugging, and design tasks traditionally central to undergraduate CS education, rendering that focus less tenable. Drawing on findings from two NUS-Google workshops in Singapore that convened faculty, industry practitioners, and students, it argues for a curriculum shift toward understanding, verifying, and co-creating with AI-generated artifacts. It identifies skills to preserve (system design, abstraction, critical evaluation) versus de-emphasize, introduces “breadcrumbs” as incremental exercises, and lists five prerequisites for reform: fostering AI-native competencies, re-centering fundamentals, enhancing advanced pathways, embracing new pedagogies, and shifting institutional support.

Significance. If the workshop-derived recommendations prove representative, the paper offers a timely, stakeholder-informed framework for adapting CS curricula to AI capabilities. The incremental “breadcrumbs” approach and explicit listing of institutional prerequisites provide concrete starting points that could influence departmental discussions and policy. The work’s strength lies in convening diverse voices (faculty, practitioners, students) rather than deriving claims from first principles or new empirical data.

major comments (2)
  1. [Abstract] Abstract (workshop description paragraph): The central recommendations for preserving system design, abstraction, and critical evaluation rest on the claim that the two workshops identified these as critical skills. However, the manuscript supplies no details on participant selection criteria, total numbers, session structure, recording methods, synthesis process, or handling of disagreement. This absence makes the specific skill list untraceable to verifiable evidence and leaves the curriculum-shift claim without a documented basis.
  2. [Findings and recommendations] Findings and recommendations section: The assertion that implementation-level programming is becoming less central is presented as following directly from AI automation, yet the text offers no counterexamples, capability benchmarks, or discussion of tasks that remain outside current generative-AI reach. Without such grounding, the premise that the traditional framing is “less tenable” cannot be evaluated against the workshops’ outputs.
minor comments (2)
  1. [Abstract] The term “breadcrumbs” is introduced in the abstract as “helpful nudges and engaging exercises” but receives no operational definition or examples in the provided text, reducing clarity for readers attempting to implement the proposal.
  2. [Proposed prerequisites] The five prerequisites for reform are enumerated without cross-references to specific workshop findings that support each item, making it difficult to assess how directly they derive from the convened discussions.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which identify opportunities to strengthen the evidential basis and transparency of the workshop-derived recommendations. We respond to each major comment below and indicate planned revisions.

read point-by-point responses

  1. Referee: [Abstract] Abstract (workshop description paragraph): The central recommendations for preserving system design, abstraction, and critical evaluation rest on the claim that the two workshops identified these as critical skills. However, the manuscript supplies no details on participant selection criteria, total numbers, session structure, recording methods, synthesis process, or handling of disagreement. This absence makes the specific skill list untraceable to verifiable evidence and leaves the curriculum-shift claim without a documented basis.

    Authors: We agree that the absence of methodological details limits traceability. The manuscript is a white paper summarizing consensus outcomes rather than a formal empirical study of the workshops. Participant selection was by invitation to NUS faculty, Google practitioners, and students; sessions used facilitated discussion rounds with iterative synthesis of themes. We will revise the abstract and add a short methods paragraph describing composition, structure, and consensus process to the extent records allow. revision: yes

  2. Referee: [Findings and recommendations] Findings and recommendations section: The assertion that implementation-level programming is becoming less central is presented as following directly from AI automation, yet the text offers no counterexamples, capability benchmarks, or discussion of tasks that remain outside current generative-AI reach. Without such grounding, the premise that the traditional framing is “less tenable” cannot be evaluated against the workshops’ outputs.

    Authors: The premise reflects workshop consensus on AI automating routine implementation. We acknowledge the lack of explicit counterexamples or limitation discussion. We will revise the findings section to include examples of tasks remaining outside current AI reach (e.g., novel architectural decisions under uncertainty, ethical oversight of AI outputs) and note current capability boundaries, providing better grounding against the workshop outputs. revision: yes

Circularity Check

0 steps flagged

No circularity; workshop synthesis is self-contained opinion reporting

full rationale

The paper reports outcomes from two NUS-Google workshops as the basis for curriculum recommendations, with no equations, fitted parameters, predictions, or derivations present. The central claims rest on participant convening and identified skills rather than any self-referential reduction, self-citation chain, or ansatz imported from prior author work. No load-bearing step reduces by construction to the paper's own inputs; the text functions as opinion synthesis without the enumerated circular patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central recommendation rests on the domain assumption that generative AI will continue to automate implementation tasks at a level that makes current curricula untenable; no free parameters or invented entities are introduced.

axioms (1)
  • domain assumption Generative AI automates many implementation-level programming, debugging, and documentation tasks
    Stated as the turning point premise in the first paragraph of the abstract.

pith-pipeline@v0.9.1-grok · 5896 in / 1160 out tokens · 28110 ms · 2026-07-01T00:52:34.748394+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 17 canonical work pages · 3 internal anchors

  1. [1]

    P . AI. Introducing perplexity computer. URL https://www.perplexity.ai/hub/blog/ introducing-perplexity-computer

  2. [2]

    Amasanti and J

    G. Amasanti and J. Jahic. The impact of ai-generated solutions on software architecture and productivity: Results from a survey study, 2025. URLhttps://arxiv.org/abs/2506.17833

    work page arXiv 2025

  3. [3]

    Bastani, O

    H. Bastani, O. Bastani, A. Sungu, H. Ge, Ö. Kabakcı, and R. Mariman. Generative ai without guardrails can harm learning: Evidence from high school mathematics.Proceedings of the National Academy of Sciences, 122(26):e2422633122, 2025

    2025

  4. [4]

    Openai claims gpt-5 model boosts chatgpt to ’phd level’

    BBC. Openai claims gpt-5 model boosts chatgpt to ’phd level’. URL https://www.bbc.com/ news/articles/cy5prvgw0r1o

  5. [5]

    Measuring the impact of early-2025 AI on experienced open-source developer productivity,

    J. Becker, N. Rush, E. Barnes, and D. Rein. Measuring the impact of early-2025 ai on experi- enced open-source developer productivity.arXiv preprint arXiv:2507.09089, 2025

    work page arXiv 2025

  6. [6]

    Brynjolfsson, B

    E. Brynjolfsson, B. Chandar, and R. Chen. Canaries in the coal mine?: Six facts about the recent employment effects of artificial intelligence. 2025

    2025

  7. [7]

    V. Charisi. Artificial intelligence in classrooms: Pedagogical dimensions. Briefing PE 784.574, European Parliament, Policy Department for Citizens, Equality and Culture, March

  8. [8]

    URL https://www.europarl.europa.eu/RegData/etudes/BRIE/2026/784574/IUST_ BRI(2026)784574_EN.pdf

    2026

  9. [9]

    Choudhuri, A

    R. Choudhuri, A. Ramakrishnan, A. Chatterjee, B. Trinkenreich, I. Steinmacher, M. Gerosa, and A. Sarma. Insights from the Frontline: GenAI Utilization Among Software Engineering Students. In2025 IEEE/ACM 37th International Conference on Software Engineering Education and Training (CSEE&T), pages 1–12. IEEE, 2025

    2025

  10. [10]

    Thinking Less, Trusting More: GenAI's Impacts on Students' Cognitive Habits

    R. Choudhuri, C. Sanchez, M. Burnett, and A. Sarma. Why johnny can’t think: Genai’s impacts on cognitive engagement.arXiv preprint arXiv:2601.22430, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  11. [11]

    Clear, A

    T. Clear, A. Cajander, A. Clear, R. McDermott, M. Daniels, M. Divitini, M. Forshaw, N. Humble, M. Kasinidou, S. Kleanthous, C. Kultur, G. Parvini, M. Polash, and T. Zhu. Ai integration in the it professional workplace: A scoping review and interview study with implications for education and professional competencies. In2024 Working Group Reports on Innova...

    work page doi:10.1145/3689187.3709607 2024

  12. [12]

    Becker, James Finnie-Ansley, Arto Hellas, Juho Leinonen, Andrew Luxton-Reilly, Brent N

    P . Denny, J. Prather, B. A. Becker, J. Finnie-Ansley, A. Hellas, J. Leinonen, A. Luxton-Reilly, B. N. Reeves, E. A. Santos, and S. Sarsa. Computing education in the era of generative ai.Commun. ACM, 67(2):56–67, Jan. 2024. ISSN 0001-0782. doi:10.1145/3624720. URL https://doi.org/10.1145/3624720. 35

    work page doi:10.1145/3624720 2024

  13. [13]

    Denny, V

    P . Denny, V. Kumar, S. MacNeil, J. Prather, and J. Leinonen. Probing the unknown: Ex- ploring student interactions with probeable problems at scale in introductory programming. InProceedings of the 30th ACM Conference on Innovation and Technology in Computer Science Education V. 1, ITiCSE 2025, page 618–624, New Y ork, NY , USA, 2025. Associa- tion for C...

    work page doi:10.1145/3724363.3729093 2025

  14. [14]

    natural language programming

    E. W. Dijkstra. On the foolishness of "natural language programming". InProgram Construction, International Summer Schoo, page 51–53, Berlin, Heidelberg, 1978. Springer-Verlag. ISBN 354009251X

    1978

  15. [15]

    Dongol, C

    B. Dongol, C. Dubois, S. Hallerstede, E. Hehner, C. Morgan, P . Müller, L. Ribeiro, A. Silva, G. Smith, and E. de Vink. On formal methods thinking in computer science education.Formal Aspects of Computing, 37(1):1–23, 2024

    2024

  16. [16]

    Eaton and S

    E. Eaton and S. L. Epstein. Artificial intelligence in the cs2023 undergraduate computer science curriculum: Rationale and challenges. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 23078–23083, 2024

    2024

  17. [17]

    Y . Fan, L. Tang, H. Le, K. Shen, S. Tan, Y . Zhao, Y . Shen, X. Li, and D. Gaševi´c. Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance.British Journal of Educational Technology, 56(2):489–530, 2025

    2025

  18. [18]

    D. Gries. What should we teach in an introductory programming course? InProceedings of the Fourth SIGCSE Technical Symposium on Computer Science Education, SIGCSE ’74, page 81–89, New Y ork, NY , USA, 1974. Association for Computing Machinery. ISBN 9781450374835. doi:10.1145/800183.810447. URLhttps://doi.org/10.1145/800183.810447

    work page doi:10.1145/800183.810447 1974

  19. [19]

    H. He, C. Miller, S. Agarwal, C. Kästner, and B. Vasilescu. Speed at the cost of quality: How Cursor AI increases short-term velocity and long-term complexity in open-source projects. In Proceedings of the 23rd International Conference on Mining Software Repositories, MSR ’26, New Y ork, NY , USA, 2026. ACM. doi:10.1145/3793302.3793349. arXiv:2511.04427

    work page doi:10.1145/3793302.3793349 2026

  20. [20]

    Scalable and Personalized Oral Assessments Using Voice AI

    P . Ipeirotis and K. Rizakos. Scalable and personalized oral assessments using voice ai, 2026. URLhttps://arxiv.org/abs/2603.18221

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  21. [21]

    Kleppmann

    M. Kleppmann. Prediction: Ai will make formal verification go mainstream, 2025. URL https: //martin.kleppmann.com/2025/12/08/ai-formal-verification.html

    2025

  22. [22]

    A. N. Kumar, R. K. Raj, S. G. Aly, M. D. Anderson, B. A. Becker, R. L. Blumenthal, E. Eaton, S. L. Epstein, M. Goldweber, P . Jalote, et al. Computer science curricula 2023, 2024

    2023

  23. [23]

    H.-P . Lee, A. Sarkar, L. Tankelevitch, I. Drosos, S. Rintel, R. Banks, and N. Wilson. The impact of generative ai on critical thinking: Self-reported reductions in cognitive effort and confidence effects from a survey of knowledge workers. InProceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pages 1–22, 2025. 36

    2025

  24. [24]

    Mohammadkarimi

    E. Mohammadkarimi. Teachers’ reflections on academic dishonesty in efl students’ writings in the era of artificial intelligence.Journal of Applied Learning & Teaching, 6(2):105–113, 2023

    2023

  25. [25]

    Moores, G

    E. Moores, G. K. Birdi, and H. E. Higson. Determinants of university students’ attendance. Educational Research, 61(4):371–387, 2019. doi:10.1080/00131881.2019.1660587. URL https: //doi.org/10.1080/00131881.2019.1660587

    work page doi:10.1080/00131881.2019.1660587 2019

  26. [26]

    S. Peng, E. Kalliamvakou, P . Cihon, and M. Demirer. The impact of ai on developer productivity: Evidence from github copilot, 2023. URLhttps://arxiv.org/abs/2302.06590

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  27. [27]

    Petrich, K

    M. Petrich, K. Wilkinson, and B. Bevan. It looks like fun, but are they learning? 2013. URL https://api.semanticscholar.org/CorpusID:111722217

    2013

  28. [28]

    Prather, P

    J. Prather, P . Denny, J. Leinonen, B. A. Becker, I. Albluwi, M. Craig, H. Keuning, N. Kiesler, T. Kohn, A. Luxton-Reilly, S. MacNeil, A. Petersen, R. Pettit, B. N. Reeves, and J. Savelka. The robots are here: Navigating the generative ai revolution in computing education. InPro- ceedings of the 2023 Working Group Reports on Innovation and Technology in C...

    work page doi:10.1145/3623762.3633499 2023

  29. [29]

    In2024 Working Group Reports on Innovation and Technology in Computer Science Education(Milan, Italy)(ITiCSE 2024)

    J. Prather, J. Leinonen, N. Kiesler, J. Gorson Benario, S. Lau, S. MacNeil, N. Norouzi, S. Opel, V. Pettit, L. Porter, B. N. Reeves, J. Savelka, I. Smith, David H., S. Strickroth, and D. Zingaro. Beyond the hype: A comprehensive review of current trends in generative ai research, teaching practices, and tools. In2024 Working Group Reports on Innovation an...

    work page doi:10.1145/3689187.3709614 2024

  30. [30]

    Denny, V

    V.-A. P˘adurean, P . Denny, A. Gotovos, and A. Singla. Prompt programming: A platform for dialogue-based computational problem solving with generative ai models. InProceedings of the 30th ACM Conference on Innovation and Technology in Computer Science Education V. 1, ITiCSE 2025, page 458–464, New Y ork, NY , USA, 2025. Association for Computing Machinery...

    work page doi:10.1145/3724363.3729094 2025

  31. [31]

    V. J. Reddi.Introduction to Machine Learning Systems. MIT Press, 2026. Available online at https://mlsysbook.ai/book/

    2026

  32. [32]

    M. I. H. Shihab, C. Hundhausen, A. Tariq, S. Haque, Y . Qiao, and B. W. Mulanda. The effects of github copilot on computing students’ programming effectiveness, efficiency, and processes in brownfield coding tasks. InProceedings of the 2025 ACM Conference on International Computing Education Research V.1, ICER ’25, page 407–420, New Y ork, NY , USA, 2025....

    work page doi:10.1145/3702652.3744219 2025

  33. [33]

    Somanath, R

    S. Somanath, R. Wakkary, O. Ettehadi, H. W. J. Lin, A. Behzad, J. Eshpeter, and D. Oogjes. Exploring the composite intentionality of 3d printers and makers in digital fabrication. 2023. URL https://api.semanticscholar.org/CorpusID:261364233

    2023

  34. [34]

    Tankelevitch, V

    L. Tankelevitch, V. Kewenig, A. Simkute, A. E. Scott, A. Sarkar, A. Sellen, and S. Rintel. The metacognitive demands and opportunities of generative AI. InCHI Conference on Human Factors in Computing Systems, pages 1–24, 2024

    2024

  35. [35]

    A. S. Theobold. Oral exams: A more meaningful assessment of students’ understanding.Journal of Statistics and Data Science Education, 29(2):156–159, 2021. doi:10.1080/26939169.2021. 1914527. URLhttps://doi.org/10.1080/26939169.2021.1914527

    work page doi:10.1080/26939169.2021 2021

  36. [36]

    Washizaki, editor.Guide to the Software Engineering Body of Knowledge (SWEBOK Guide)

    H. Washizaki, editor.Guide to the Software Engineering Body of Knowledge (SWEBOK Guide). IEEE Computer Society, 4.0 edition, 2024. URLhttps://www.swebok.org

    2024

  37. [37]

    L. Yu. Paradigm shift on coding productivity using genai. InProceedings of the 29th International Conference on Evaluation and Assessment in Software Engineering, EASE ’25, page 708–713, New Y ork, NY , USA, 2025. Association for Computing Machinery. ISBN 9798400713859. doi: 10.1145/3756681.3757081. URLhttps://doi.org/10.1145/3756681.3757081. 38

    work page doi:10.1145/3756681.3757081 2025